Abstract Background Perioperative hemoglobin drop is widely used as a surrogate marker for blood loss and transfusion risk across surgical disciplines. However, its agreement with dilution-based estimates of hidden blood loss remains insufficiently validated. As both approaches represent indirect estimators rather than a reference standard, this study evaluates agreement rather than validation. Methods This retrospective registry-based study included 828 patients undergoing elective primary total hip arthroplasty between 2016 and 2023. Hemoglobin drop was defined as the difference between preoperative hemoglobin and the lowest postoperative value within 48 h. Hidden blood loss was calculated using the Gross–Sehat method with blood volume estimation according to Nadler. Agreement between hemoglobin drop and hidden blood loss was assessed using Bland–Altman analysis and intraclass correlation coefficients. Discriminative performance for identifying patients with high hidden blood loss (upper quartile) was evaluated using receiver operating characteristic analysis. Results Mean perioperative hemoglobin drop was 33.5 ± 12.3 g/L. Mean hidden blood loss was 854.9 mL, accounting for the majority of total blood loss. Agreement between hemoglobin drop and hidden blood loss was limited, with wide limits of agreement on Bland–Altman analysis. Although a positive association was observed, substantial interindividual variability persisted, particularly at higher blood loss levels. Hemoglobin drop demonstrated only moderate discriminative performance for identifying patients with high hidden blood loss. Conclusion Perioperative hemoglobin drop shows only moderate agreement with dilution-based estimates of hidden blood loss and cannot reliably replace formal blood loss calculations at the individual patient level. These findings have implications for perioperative blood management and caution against reliance on hemoglobin change alone for blood loss assessment. Level of Evidence (LoE) Level III - retrospective cohort study / registry-based observational study.
Ramadanov et al. (Wed,) studied this question.